[08:35:50] <gehel>	 Errand, back in a few 
[10:09:26] <dcausse>	 lunch
[11:37:31] <pfischer>	 lunch
[16:15:52] <ebernhardson>	 hmm, so in k8s i can see the namespace in `kubectl get namespaces`, using `helmfile -e staging diff` doesn't report anything, which implies that it's all deployed, but `kubectl get releases -n cirrus-streaming-updater` gives nothing :S
[16:17:09] <dcausse>	 ebernhardson: look at kubectl get flinkdeployment
[16:17:15] <dcausse>	 I see both apps
[16:17:37] <dcausse>	 but they're not taken care of by the operator apparently
[16:18:08] <dcausse>	 so either the operator is not listening to this namespace or it's not liking something
[16:21:28] <ebernhardson>	 hmm
[16:22:45] <ebernhardson>	 will try and find logs for the flink operator i suppose
[16:23:13] <dcausse>	 https://logstash.wikimedia.org/app/dashboards#/view/f3fefa60-f95a-11ed-aacf-e115c4d3fd2c?_g=h@543b7f0&_a=h@794b964
[16:23:33] <dcausse>	 had a quick look but could not find anything
[16:24:02] <dcausse>	 might be worth trying to helmfile destroy and then apply and have the logs open at the same time to see what moves
[16:24:38] <ebernhardson>	 sounds reasonabl
[16:25:19] <dcausse>	 also I'm not sure if staging is eqiad or codfw after the switch
[16:25:31] <ebernhardson>	 according to symlinks it's still eqiad, but who knows :)
[16:25:53] <ebernhardson>	 i mean /etc/helmfile-defaults/general-staging.yaml points to general-staging-eqiad.yaml
[16:25:58] <dcausse>	 ok
[16:26:30] <ebernhardson>	 i had tried to directly put staging-eqiad in the chart, seemed sensible, but the test suite only checks 'staging' and not staging-eqiad
[16:27:35] <dcausse>	 yes I barely remember that they want to keep one staging k8s cluster free of deployments to test k8s specific things
[16:30:29] <ebernhardson>	 ahh, i suppose that makes sense
[16:31:35] <dcausse>	 ebernhardson: I think your namespace is missing here: https://gerrit.wikimedia.org/r/plugins/gitiles/operations/deployment-charts/+/refs/heads/master/helmfile.d/admin_ng/values/staging-eqiad/flink-operator-values.yaml#6
[16:31:53] <ebernhardson>	 dcausse: oh! yea that would make sense
[16:32:40] <ebernhardson>	 i wonder if that can be inverted somehow, i would think we would rather flag namespaces than maintain a central list.  But for another day :P
[16:32:53] <ebernhardson>	 (also there may be no such concept as flagging/tagging namespaces :P)
[16:35:14] <ebernhardson>	 ryankemper: if you're arround, could use review/merge on https://gerrit.wikimedia.org/r/c/operations/deployment-charts/+/964567
[16:37:59] <dcausse>	 yes it's sad that the operator has to be aware of the namespaces to listen to, would be nice if there was some kind of annotations that could be used to filter those, like how prometheus uses "prometheus.io/scrape": "true"
[16:38:24] <Trey314159>	 FYI—ebernhardson dr0ptp4kt inflatador—today is a US holiday. Oops. I'm going to take the next 24 hours off, I guess. See you all tomorrow.
[16:39:14] <ebernhardson>	 hmm, well ok :)
[16:39:49] <dr0ptp4kt>	 thanks Trey3.14159 . i'm looking for a day to which to flex, although my my the weather outside is good today
[16:40:45] <Trey314159>	 I usually notice the "non-celebratory" US holidays but I missed this one somehow